Skip to content
llms.py logo llms.py

Web UI

llms.py includes a lightweight, fast, and privacy-focused web UI that provides a ChatGPT-like experience for accessing all your configured LLM providers.

Starting the UI

Section titled “Starting the UI”
Terminal window
llms --serve 8000

This launches:

  • Web UI at http://localhost:8000
  • OpenAI-compatible API at http://localhost:8000/v1/chat/completions

Key Features

Section titled “Key Features”

🔒 Privacy First

Section titled “🔒 Privacy First”
  • Offline Operation: Works entirely offline once loaded
  • Local Storage: All data stored in browser’s IndexedDB
  • No Tracking: No analytics, ads, or external dependencies
  • No Sign-ups: Free and open source

⚡ Fast & Lightweight

Section titled “⚡ Fast & Lightweight”
  • Zero Build Tools: Pure JavaScript modules
  • No npm Dependencies: Simple, modern JavaScript
  • Fast Loading: Minimal bundle size
  • Responsive: Works on desktop and mobile

🎨 Modern Interface

Section titled “🎨 Modern Interface”
  • Dark Mode: Automatic or manual toggle
  • Markdown Support: Full markdown rendering
  • Syntax Highlighting: Code blocks with copy buttons
  • Responsive Layout: Collapsible sidebar

Core Features

Section titled “Core Features”

Chat Interface

Section titled “Chat Interface”
  • Multi-turn Conversations: Maintain context across messages
  • Edit Messages: Modify and re-run previous prompts
  • Copy Responses: Quick copy for messages and code blocks
  • Thread Management: Organize conversations in threads

Model Selection

Section titled “Model Selection”
  • Smart Autocomplete: Quick model search and selection
  • Provider Filtering: Only shows models from enabled providers
  • Cost Display: See input/output pricing per 1M tokens
  • Real-time Updates: Models update when providers are enabled/disabled

System Prompts

Section titled “System Prompts”
  • 200+ Built-in Prompts: Professional prompts for various use cases
  • Quick Search: Autocomplete for finding prompts
  • Custom Prompts: Add your own in ui.json
  • Categories: Organized by use case

Multi-Modal Support

Section titled “Multi-Modal Support”
  • Image Upload: Drag & drop or paste images
  • Audio Upload: Process audio files
  • PDF Attachments: Upload and analyze documents
  • Multiple Formats: Support for various file types

Analytics & Metrics

Section titled “Analytics & Metrics”

Thread-Level Metrics

Section titled “Thread-Level Metrics”

Each conversation thread displays:

  • Total cost
  • Token count (input/output)
  • Number of requests
  • Total response time

Message-Level Metrics

Section titled “Message-Level Metrics”

Individual messages show:

  • Token count
  • Cost (for premium models)
  • Response time
  • Model used

Analytics Pages

Section titled “Analytics Pages”

Cost Analytics

Section titled “Cost Analytics”
  • Daily cost breakdown by month
  • Per-model and per-provider costs
  • Interactive charts
  • Expandable daily details

Token Analytics

Section titled “Token Analytics”
  • Daily token usage by month
  • Input/output token breakdown
  • Per-model and per-provider usage
  • Visual charts

Activity Log

Section titled “Activity Log”
  • Detailed request history
  • Model, provider, and prompt info
  • Token counts and costs
  • Response times and speeds
  • Searchable and filterable

Data Management

Section titled “Data Management”

Import/Export

Section titled “Import/Export”
  • Export Chats: Backup all conversations
  • Import Chats: Restore from backup
  • Export Logs: Backup analytics data (hold ALT while clicking Export)
  • Import Logs: Restore analytics data

Search History

Section titled “Search History”
  • Full-text Search: Find past conversations
  • Real-time Results: Instant search as you type
  • Context Preview: See matching messages

Provider Management

Section titled “Provider Management”

Enable/Disable Providers

Section titled “Enable/Disable Providers”

Toggle providers on/off directly in the UI:

  • Click provider toggle next to model selector
  • Changes persist to configuration
  • Models update immediately

Provider Order

Section titled “Provider Order”

Providers are tried in the order defined in llms.json:

  1. Free tier providers first
  2. Local providers (Ollama)
  3. Premium cloud providers

Advanced Settings

Section titled “Advanced Settings”

Access advanced options through the settings dialog:

  • Temperature (0-2): Control randomness
  • Max Tokens: Limit response length
  • Top P (0-1): Nucleus sampling
  • Frequency Penalty (-2 to 2): Reduce repetition
  • Presence Penalty (-2 to 2): Encourage diversity
  • Stop Sequences: Control generation stopping
  • Seed: Reproducible outputs
  • Reasoning Effort: For reasoning models
  • And more: Additional model-specific parameters

Keyboard Shortcuts

Section titled “Keyboard Shortcuts”
  • Enter: Send message
  • Shift+Enter: New line in message
  • Esc: Close dialogs/selectors

Configuration

Section titled “Configuration”

UI settings are stored in ~/.llms/ui.json:

{
  "prompts": [
    {
      "id": "it-expert",
      "name": "Act as an IT Expert",
      "value": "I want you to act as an IT expert..."
    }
  ]
}

Next Steps

Section titled “Next Steps”